|About this Abstract
||2022 TMS Annual Meeting & Exhibition
||High Performance Steels
||Understanding Deformation-induced Cracking in Dual-phase Steel via the Combination of EBSD Analysis and Convolutional Neural Network
||Hung-Wei Yen, Po-Hsun Lin, Yi-Fan Hu, Min-Yu Tseng, Kuo-Cheng Yang, Kangying Zhu
|On-Site Speaker (Planned)
Dual-phase (DP) steel has been widely used in vehicles due to its excellent strength-ductility balance. However, micro-cracks could be induced during cold forming, leading to unqualified components. The study on formation of micro-crack in DP steel is a long history and primary factors are attributed to chemical composition, loading mode, strength grade, and phase constituent. Until today, the principles of micro-cracking are not yet unified.
In this work, we developed a state-of-art algorithm by convolution neural network to predict crack formation from EBSD analysis. Machine learning and analyses were done for DP steels from 590 MPa grade to 1180 MPa grade. The essentials of cracking sites for DP steels of various grades are classified based on phase constituent. Moreover, we developed an approach to predict crack propagation from the learning results. The research will contribute on the microstructural design to optimize formability of DP steels.
||Iron and Steel, Machine Learning, Mechanical Properties